Inherited thrombophilia is common in the Caucasian population with a prevalence of up to 15% (Greer, 2003). Unsurprisingly, therefore, these abnormalities are commonly found in women with RPL but this does not prove causation.
Interest in the role of inherited thrombophilia in this setting has grown since the publication in 1996 of a study examining the risk of fetal loss in women with familial thrombophilia enrolled in the European Prospective Cohort on Thrombophilia (EPCOT) study (Preston et al, 1996). The overall risk of pregnancy loss was increased in 571 women with thrombophilia with an odds ratio (OR) of 1·35 [95% confidence interval 1·01–1·82]. The OR was higher for stillbirth (3·6 [1·4–9·4]) than for miscarriage (1·27 [0·94–1·71]) and within this late fetal loss group (after 28 weeks of gestation) the OR was highest in women with combined defects (14·3 [2·4–86·0]), followed by antithrombin deficiency (5·2 [1·5–18·1]), protein C deficiency (2·3 [0·6–8·3]), protein S deficiency (3·3 [1·0–11·3]) and factor V Leiden (F5 R506Q) (2·0 [0·5–7·7]). In the miscarriage groups (losses occurring at or prior to 28 weeks of gestation) the corresponding ORs for these inherited thrombophilias were 0·8 (0·3–2·6), 1·7 (1·0–2·8), 1·4 (0·9–2·2), 1·2 (0·7–1·9) and 0·9 (0·5–1·5) respectively. With such wide confidence intervals, the majority of which cross the boundary of 1·0, overall these data are unconvincing for an association between inherited thrombophilia and pregnancy loss.
The TREATS (Thrombosis: Risk and Economic Assessment of Thrombophilia Screening) Study was a systematic review of thrombophilia in pregnancy that included a total of 79 studies; three randomized controlled trials, eight prospective cohorts and 68 retrospective studies (Robertson et al, 2006). They examined the association between thrombophilia and early and late pregnancy loss separately. Twenty‐five studies assessed thrombophilia in early losses (defined as recurrent first or single second trimester); significant associations were observed with homozygous F5 R506Q (2·71 [1·32–5·58]), heterozygous F5 R506Q (1·68 [1·09–2·58), F2 (prothrombin gene) mutation heterozygosity (2·49 [1·24–5·00]), anticardiolipin antibodies (3·40 [1·33–8·68]), lupus anticoagulant (LA) (2·97 [1·03–8·56]), acquired activated protein C resistance (4·04 [1·67–9·76]) and hyperhomocysteinaemia (6·25 [1·37–28·42]). Analysis of the F5 R506Q data combined both homozygotes and heterozygotes and found a higher risk of pregnancy loss in the second (4·12 [1·93–8·81]) compared to the first trimester (1·91 [1·01–3·61]). Similar to F5 R506Q, prothrombin mutation heterozygosity also increased the risk of recurrent first trimester losses and non‐recurrent second trimester loss by OR 2·70 [1·37–5·35] and 8·6 [2·18–33·95], respectively, with an association being most convincing for the latter. Fifteen studies related to late pregnancy loss (defined as the third trimester) and showed an increased risk in F5 R506Q heterozygotes (2·06 [1·10–3·86]), F2mutation heterozygotes (2·66 [1·28–5·53]), protein S deficiency (20·09 [3·7–109·15]) and anticardiolipin antibodies (3·30 [1·62–6·70]). While these pooled data are more suggestive of an association, the wide confidence intervals, with the lower boundary frequently close to one, and the heterogeneity of the data, particularly pertaining to F5 R506Q and recurrent first trimester losses (χ2 = 23·66, df = 7, P = 0·001), again signify the need for caution in interpretation.
A modest association was evident in another meta‐analysis, which included a total of 31 studies (of which only two were prospective cohorts) and again examined associations according to the timing of the pregnancy losses (Rey et al, 2003). F5 R506Q was associated with early RPL (OR 2·01 [1·13–3·58]), late RPL (7·83 [2·83–21·67]) and late non‐recurrent fetal loss (3·26 [1·82–5·83]); F2 mutation with early RPL (2·56 [1·04–6·29]) and late non‐recurrent loss (2·3 [1·09–4·87]); protein S deficiency was associated with late non‐recurrent fetal loss (7·39 [1·28–42·63]); protein C and antithrombin deficiency were not significantly associated with any type of pregnancy loss. Due to the lower prevalence of women with natural anticoagulant deficiencies (antithrombin, protein C and protein S) there are far fewer studies investigating their associations with adverse obstetric outcomes and data thus far is based on low patient numbers. Without statements to the contrary, it is possible that some women in these studies received antenatal venous thromboembolism (VTE) prophylaxis. In turn, if the mechanism of pregnancy loss in these patients is due to thrombosis in uteroplacental vessels it is possible that the prophylactic therapy may have contributed to the success of the pregnancy and therefore any association may be missed.
Thrombophilia and the pathophysiology of pregnancy loss
Pregnancy loss in APS has traditionally been ascribed to uteroplacental thrombosis and was first considered after the finding of massive placental infarction in a lupus anticoagulant (LA) positive woman who experienced an intrauterine death (De Wolf et al, 1982). Similar findings from women with antiphospholipid (aPL) antibodies and pregnancy loss supported this theory (Out et al, 1991). Recently there has been more interest in the role of the trophoblast and endometrial invasion and implantation, with focus on an underlying immunomodulatory, rather than purely thrombotic process. Histopathological examination of products of conception in women with primary APS and pregnancy losses between 7 and 10 weeks' gestation attributed these losses to abnormal endovascular trophoblast invasion in decidual vessels rather than excessive intervillous thrombosis (Sebire et al, 2002). Complement activation has previously been reported as causative in aPL antibody‐induced fetal injury, with suggestions that heparin was beneficial due to anti‐complement rather than anticoagulatory effects (Girardi et al, 2004). Murine experiments supported this theory as both unfractionated heparin (UFH) and the low molecular weight heparin (LMWH) enoxaparin (even at sub‐therapeutic dosage) demonstrated an inhibitory effect on complement activation and protected mice from aPL antibody‐induced pregnancy complications; neither fondaparinux (a specific inhibitor of clotting factor Xa) nor hirudin (a direct thrombin inhibitor) had either effect (Girardi et al, 2004). A third proposed mechanism of pregnancy loss in women with APS concerns annexin V. Previously known as placental anticoagulant protein 1, annexin V is produced by villous trophoblasts and has potent anticoagulant activity due to a high affinity for anionic phospholipids (Rand et al, 1997). Clustering of annexin V on phospholipid surfaces results in displacement of clotting factor Va, precluding formation of procoagulant complexes (Andree et al, 1992). Thus, removal of annexin V from trophoblast membranes (for example by anti‐annexin V or aPL antibodies) induces a procoagulant surface; markedly reduced levels on placental villi have been demonstrated in women with APS (Krikun et al, 1994; Rand et al, 1994).
Another mechanism is suggested by murine studies, which demonstrated that passive transfer of a human monoclonal antiphospholipid antibody, CIC15, isolated from a patient with primary APS and recurrent early pregnancy losses induced fetal resorption (Lieby et al, 2004). Histological analysis revealed signs of decidual arterial thrombosis, but there was no evidence of inflammatory cell infiltration in the decidual or fetal tissue. CIC15 was unable to disrupt the annexin V shield (unlike other aPL antibodies), suggesting that pregnancy loss was neither due to displacement of annexin V from trophoblast surfaces nor inflammation. Although the precise pathogenicity remains to be identified, in vitro experiments support the idea that pregnancy loss in this setting was probably related to the procoagulant activity of CIC15 (Poindron et al, 2011). Analogous to the earlier heparin experiments (Girardi et al, 2004), this more recent work demonstrated that LMWH (tinzaparin at a therapeutic dose) completely protected mice from fetal injury induced by CIC15. Both fondaparinux and hirudin were also protective, suggesting that CIC15‐mediated fetal injury is largely a consequence of a prothrombotic effect